VII. Mathematics in the Third Millennium?
[Based on my talk on ``Mathematics in the third millennium'' at Tor
Nørretrander's fabulous Mindship institute, Copenhagen, summer of
1996. Also based on the interview with me conducted by Guillermo
Martínez and published June 1998 in the Buenos Aires newspaper
Página/12.]
Synopsis
Is math quasi-empirical? (Again!)
I don't believe there should be an abrupt discontinuity between
how mathematicians work and mathematical physicists work—it
should be a continuum of possibilities.
Randomness & entropy in physics versus lack of structure
defined via program-size complexity:
Boltzmann, individuals versus ensembles.
Wolfram: maybe the universe is like π, pseudo-random!
Mathematical discovery:
Discovery versus formal reasoning, Euler versus Gauss,
Polya's Mathematics and Plausible Reasoning,
reading Euler's Opera Omnia as a child!
Biological complexity, evolution & the origin of life!?
My complexity is too hard to increase. Wolfram:
because of the ubiquity of universality, maybe evolution
is easy!
Nature is a cobbler, un bricoleur.
Contrast biology with elementary number theory.
Material from Guillermo Martínez interview. Beauty of
mathematics, simple, powerful, elegant ideas. Is math becoming like
biology, messy, complex?
Complicated contemporary physics. No simple equations, no hydrogen
atom. Now it's many-body statistical physics. Even fundamental
physical theory is like that: the quantum field vacuum is a hot bed of
activity. Joke from book on many-body problems on the progress of
physics, i.e., how many bodies does it take to have a problem?
The Beauty of Mathematics
When I was young, one of the things that attracted me a great deal was
the beauty of mathematics. I had similar
feelings when I read and understood a beautiful mathematical idea as
when I saw a beautiful painting, a beautiful woman,
or a graceful ballerina. Human
society might be a mess, life a chaotic tragedy, but I could escape into the
beautiful, clear, sharp, inhuman light of elementary number theory, of
the prime numbers, where a few simple powerful elegant ideas were what
counted, not power, not violence, not money!
I recall that at school I was good at subjects that required
reasoning, not memorization. I was very good at math and physics;
everything could be deduced from the basic principles.
I was bad at French; there were not enough simple powerful
unifying principles.
Take a good look at biology! It's a complicated mess. Are there laws
of biology in the same sense as there are laws of physics? Nature is
a cobbler, nature patches and reworks biological organisms, they're a
mess, but they work, they survive! That's natural selection for you!
I liked physics as well as math. Look at the Bohr model for the
hydrogen atom, or at the Schrödinger equation. A few simple
equations explained it all!
Well, a funny thing has been happening. Math has been getting more
complicated. Look at the immense computer proof-by-cases for the
four-color theorem.
[That's the assertion that with four colors you can paint any map on the plane
in such a way that adjacent countries have different colors.]
Look at the human generated but still monstrous
classification of all simple groups: ten-thousand pages of proofs
written by many, many mathematicians!
[Roughly speaking, simple groups play the same role in group theory
that the primes play in number theory.
For understandable explanations of the proof of the four-color
theorem and of the classification of the simple groups, see
L.A. Steen, Mathematics Today—Twelve Informal Essays.]
And look at contemporary physics. Now you don't do theoretical physics
writing down a simple equation and solving it analytically in closed
form, like you did when I was a child. Now it's complicated computer
models that you simulate on the computer to see how they behave...
[See for example G.W. Flake, The Computational Beauty of
Nature—Computer Explorations of Fractals, Chaos, Complex Systems,
and Adaptation.]
Complicated contemporary physics! No simple equations, no hydrogen
atom. Now it's many-body statistical physics. Even fundamental
physical theory is like that. Even the quantum field vacuum is a
hotbed of activity. Here's a joke, from a book on many-body problems,
on the progress of physics, as measured by how many bodies it takes to
have a problem.
Richard Mattuck,
in his book he modestly refers to as Feynman Diagrams for Idiots (the
official title is A Guide to Feynman Diagrams in the Many-Body
Problem), sums up the progress of physics like
this. How many bodies does it take to have a problem? In Newtonian
physics, it was three bodies. Two gravitating point masses you can
solve exactly in closed form, three, no. In general relativity two
bodies get you in trouble. For a single mass point, you have the neat
Schwarzchild solution, also known as the black
hole. But for two bodies, it's complicated numerical work on the
computer... Now in quantum field theory, even zero bodies is too much!
Because the quantum mechanical vacuum is very complicated, it's a
seething sea of creation and annihilation of virtual particles... You
can do perturbation expansions to make estimates, but exact
closed-form solutions? Forget it!
[For an understandable explanation of quantum field theory, see
R.P. Feynman, QED—The Strange Theory of Light and Matter...
Let me give a particularly dramatic—but more technical—example.
Following what's called the lattice gauge theory approach,
my colleague Don Weingarten built a massively parallel super-computer
just in order to do Monte Carlo estimates (estimates via statistical sampling)
of Feynman path integrals (sums over all histories) in QCD (quantum
chromodynamics, the theory of quarks and gluons).
Each computation took about a year!]
So what will the mathematics of the future be like? Will there be
wonderful new simple powerful ideas, or will things be messy and
complicated as in biology? In that case, new kinds of scientific
personalities will be needed to do this new kind of mathematics...
Well, it's a funny thing, but if you look back at the work of
Gödel, Turing, and my own that I've presented here, this was
already happening... You can already clearly see the beginning of a new
kind of mathematics, a very different kind of mathematics, one that is
more complicated, one that is in a way more like biology...
Here's why I say this...
A new complicated mathematics?
My approach is, in a way, just as complicated as Gödel's and
Turing's, except that the complications are different. For
Gödel, it's the internal structure of his axiomatic system and
his primitive recursive definitional schemes and his Gödel
numbering that's complicated. For Turing, it's the universal Turing
machine interpreter program, which he spells out in his 1936 paper.
And for me it's the LISP interpreter (which corresponds to Turing's
complicated universal machine), which you don't see, and the
definition of the LISP language, the size of the programmer's manual,
which you do see. In my case, the complications are like an
iceberg, most of which is below the water!
Peano arithmetic + first order logic as a formal axiomatic system, the
code for Turing's universal machine, the interpreter for my LISP...
These are very strange kinds of mathematical objects, completely
different from traditional mathematical objects... Look at the primes,
at the Riemann ζ function ζ(s), they're so simple... Look at a workable
formal axiomatic system, at Turing's universal machine, at a LISP
interpreter, they're so complicated...
So in a way, in all three cases, Gödel, Turing, and I, we already
have a new ``biological'' complicated mathematics, the mathematics of the
third millennium, or at least of the 21st century.
[As a child I used to dream that I was in the far future, in a library,
desperate to see how it had all turned out, desperate to see what
science had achieved. And I would take a volume off the shelf and open
it, and all I could see were words, words, words, words that made no sense at all...
Writing this book brings back long-forgotten thoughts
and the unusual lucidity I experience when my
research is going well and everything seems inevitable.]
Integrative themes: Information, Complexity, Randomness
In a way, these three words really sum up and tie together an immense complicated
scientific and technological paradigm shift at the end of this century
and of this millennium. They sum up the new zeitgeist, the new spirit
of these times.
Look at DNA, it's biological information... Look at the new field of
quantum computing and quantum information theory... Look at the title
of my colleague Rolf Landauer's 1991 paper in Physics Today:
``Information is physical''...
Look at how complicated computer hardware and especially software is
becoming... At the megabytes and megabytes of code one is now
accustomed to have, and that you need to have, to use a
computer...
Look at the human genome project, it's so much information, a huge
data base of it, many huge databases... And one needs new software
technology to organize it, to search it, to use it...
[See D.S. Robertson, The New Renaissance—Computers and the Next
Level of Civilization, for a deep information-theoretic analysis of
the four different levels of civilization associated with speech, reading
and writing, the printing press, and the PC, the Internet and the Web.
According to Robertson, the key feature of each of these steps forward
has been a substantial increase in the amount of information that can
be stored, remembered and processed by the human race. And each of
these jumps in information-processing power is associated with major
social change.]
Look at artificial intelligence. I think it's happening, I think
we're half-way there, we just don't realize it. People used to think,
AI pioneers used to think, that they just needed a handful of great
ideas, Nobel-prize-winning level ideas, and they would understand
how human intelligence works and how to create an artificial
intelligence. Instead we're getting chess playing, speech recognition
and synthesis, etc., by accretion, by summing the work on hardware and
software by an entire planet of hardware and software engineers...
It's not a few fundamental new ideas, it's megabytes and megabytes of
complicated software, that is gradually developing and evolving...
Look at some recent speculations on the nature of consciousness
[D.J. Chalmers, The Conscious Mind—In Search of
a Fundamental Theory,
G.R. Mulhauser, Mind Out of Matter—Topics in the Physical
Foundations of Consciousness and Cognition,
T. Nørretranders, The User
Illusion—Cutting Consciousness Down to Size]
where information theory is discussed. Consciousness does not seem to
be material, and information is certainly immaterial, so perhaps
consciousness, and perhaps even the soul, is sculpted in information,
not matter.
As science fiction writers are fond of pointing out, ``soul'' is to
``body'' as ``program'' is to ``computer.''
The conventional view is that matter is primary, and that
information, if it exists, emerges from matter. But what if
information is primary, and matter is the secondary phenomenon!
After all, the same information can have many different material
representations in biology, in physics, and in psychology: DNA, RNA;
DVD's, videotapes; long-term memory, short-term memory, nerve impulses, hormones.
The material representation is irrelevant, what counts is the
information itself.
The same software can run on many machines.
Information is a really revolutionary new kind of concept,
and recognition of this fact is one of the milestones of this
age.
That really sums up what I have to say, what I see as the moral of the
story... What I see as the broad picture... But I can't resist a few
more detailed final remarks... Some final words...
Afterthoughts...
What is Ω? It's just the diamond-hard distilled and
crystallized essence of mathematical truth! It's what you get when you
compress tremendously the coal of redundant mathematical truth... And
is math quasi-empirical? (Not that again!) Let me state my position
as modestly and uncontroversially as possible: I don't believe there
should be an abrupt discontinuity between how mathematicians work and
mathematical physicists work—it should be a continuum of
possibilities. No proof is totally convincing. There are just
differing degrees of credibility.
[I am not saying that math and physics are one and the same;
math deals with the world of mathematical ideas and physics deals with
the real world, math is quasi-empirical and physics is empirical.
In particular there is a big difference between the two subjects that
was drummed into me when I was a guest in Gordon Lasher's theoretical
physics group. That's the fact that physicists know that no equation
is exact—they're merely good approximations in which one ignores
lower-order effects, in which one ignores perturbations that operate on
smaller scales. As Jacob Schwartz so beautifully put it in an essay in
M. Kac, G.-C. Rota, and J.T. Schwartz's anthology Discrete
Thoughts—Essays on Mathematics, Science, and Philosophy, physicists
know that all equations are approximate, so they prefer short, robust,
unrigorous proofs that are stable under perturbations, to long, fragile,
rigorous proofs that are not stable under perturbations (but that are
perfectly okay in pure mathematics)... I also strongly recommend Gian-Carlo
Rota's anthology Indiscrete Thoughts. Among his other fascinating
observations on doing mathematics, Rota makes the point that some
mathematicians are mental athletes who like finding new proofs and settling
old problems, while others are dreamers who prefer to find new definitions
and create new theories. I definitely belong to the latter class!
Rota makes the point that these two extremely different kinds of
mathematical personalities sometimes view each other with thinly veiled
contempt!... By the way, these remarks cost Rota some friends.
And that's another difference between mathematics and physics:
physicists have a sense of humor, mathematicians don't!]
I should state here that AIT has an intimate connection with physics.
Charles Bennett and others have used program-size instead of Boltzmann
entropy in their discussion of Maxwell's demon.
Two very readable books on this subject were published in 1998. See
T. Nørretranders, ``Maxwell's Demon,'' chapter 1 in The User
Illusion—Cutting Consciousness Down to Size, and
H.C. von Baeyer, Maxwell's Demon—Why Warmth Disperses and
Time Passes.
Let's compare randomness and entropy in physics
with lack of structure as defined via program-size complexity. It's
just individuals versus ensembles! In statistical physics you have
Boltzmann entropy which measures how well probability is distributed
over an ensemble of possibilities. It's an ensemble notion. In
effect, in AIT I look at the entropy/program-size of individual
microstates, not at the ensemble of all possible microstates and the
distribution of probability across the phase space. For more on the
history of these ideas, see David Ruelle's delightful
book Chance and Chaos.
Some final words on Stephen Wolfram's fascinating, and unfortunately
unpublished, ideas. Wolfram has a very different view of complexity
from mine. In my view π is not at all complex, but to Wolfram
it's infinitely complex, because it looks completely random.
Wolfram's view is that simple laws, simple combinatorial structures,
can produce very complicated unpredictable behavior. π is a
good example. If you didn't know where they come from, its digits
would look completely random. In fact, Wolfram says, maybe the
universe contains no randomness, maybe everything is actually
deterministic, maybe it's only pseudo-randomness! And how could
you tell the difference?
The illusion of free will is because the future is too hard to
predict, but it's not really unpredictable.
[To Wolfram's exceedingly bright and sharp mind, the idea of
indeterminacy, of randomness, of something irrational, that escapes
the power of reason, of simple unifying principles, that happens for
no reason—and that he will never be able to
understand—is totally abhorrent. The horror of a vacuum of the
ancients becomes a modern horror of randomness. To such a mind, I
must appear, because of my belief in randomness, as a muddle-headed
mystic!... I'm also reminded of Feynman's fury in a conversation we
had near the end of his life when I suggested that there might be
wonderful new laws of physics waiting to be discovered. Of course!, I
told myself later, how could he bear the thought that he wouldn't live
to see it?... Science and magic both share the belief that ordinary
reality is not the real reality, that something more fundamental is
hidden behind everyday appearances. They share a belief in the
fundamental importance of hidden secret knowledge. Physicists are
searching for their TOE, theory of everything, and kabbalists search for
a secret name of God that is the key that unlocks all understanding.
In a way the two are allies, for neither can bear the thought that
there is no secret meaning, no final theory, and that things may be
arbitrary, random, meaningless, incompressible and incomprehensible.
For a dramatization of this idea, see D. Aronofsky's 1998 film π.
See also G. Johnson, Fire in the Mind—Science, Faith, and the
Search for Order, and P. Davies, The Mind of God—The Scientific
Basis for a Rational World.]
Wolfram also has some fascinating ideas about biology, the origin of
life and evolution. One of my big disappointments, the big
disappointment in my scientific life, is that I couldn't use my
program-size complexity to make a mathematical theory out of Darwin.
[I was strongly influenced by von Neumann. For an early report of von Neumann's
ideas, see J.G. Kemeny's 1955 article in Scientific American,
``Man viewed as a machine.'' For a statement by von Neumann himself,
see ``The general and logical theory of automata'' in volume 4 of
J.R. Newman's The World of Mathematics. For a posthumous
account assembled by A.W. Burks, see von Neumann's Theory of Self-Reproducing
Automata. For samples of contemporary thought on these matters,
see P. Davies, The Fifth Miracle—The Search for the Origin of Life,
and C. Adami, Introduction to Artificial Life.]
My complexity is conserved, it's impossible to make it increase, which
is great if you're doing metamathematical incompleteness results, but
hell if you want to get evolution. So I asked Wolfram his thoughts on
this matter, and his reply was absolutely fascinating. He has amassed
much evidence of the ubiquity of universality.
In other words, he's discovered that many,
many different kinds of simple combinatorial systems achieve
computational universality, and have rich, complicated unpredictable
behavior. π is just one example... So what's so surprising
about getting life, about getting clever organisms that exhibit rich,
complicated behavior, that need it to survive? That's easy to
do!!! And I suspect that Wolfram is right, I just want to get a copy
of his 800-page book on the subject and be able to read it and think
about it at my leisure. I have held its two volumes in my hands,
briefly, once, during a fascinating visit to Wolfram's home...
A final, very final, word on mathematical discovery. It's been fun,
great fun, for me to work on incompleteness and information. But
incompleteness results are depressing, and formal systems are a drag,
a bore... It's much more fun to think about mathematical discovery,
about creativity instead of formal reasoning, and about L. Euler instead of
C.F. Gauss. Why Euler versus Gauss? Because Euler published every step in
his reasoning, in the discovery process, while Gauss carefully removed
all the scaffolding from around his beautiful buildings... Gauss's papers,
I'm told, are very hard to read... P.G.L. Dirichlet traveled with
Gauss's masterpiece
Disquisitiones Arithmeticae everywhere for years. But Euler is
a delight to read...
I still remember my childish joy at reading the story of how Euler
made some of his great mathematical discoveries in Polya's two volume
Mathematics and Plausible Reasoning. As a child I was lucky
enough to get permission to wander through the stacks at the Columbia
University mathematics library, and I was fascinated by some of the
collected works that I found there: N.H. Abel's collected works
(Oeuvres Complètes), which are small but wonderful, and in
beautiful old French, and Euler's collected works (Opera
Omnia), which are anything but small! They're still being
gradually published; Euler left so many manuscripts...
I remember what
a joy it was to read a series of papers on number theory by Euler and
see the evidence that led him to conjecture a
result, and how he gradually filled in the holes in a proof
until he finally had a complete proof! What a
treat it was for me to translate one of his number theory papers
written in Latin... I knew no Latin, I just had a Latin
dictionary—but I did know plenty of number theory! Or his paper in
French explaining his discovery of a recursion formula for
σ(n), the sum of the divisors of the natural number n,
what a delight!
So no more depressing incompleteness results!
No more cold, dry formal axiomatic systems! A sensual, joyful
theory of discovery, of creation, that's what I want! My theorems may
be pessimistic, but I'm an optimist!
[For more evidence of this, see my interview in J. Horgan, The End
of Science—Facing the Limits of Knowledge in the Twilight of the
Scientific Age.]
Maybe you or some other reader
of this book can find a way to do it! After all, all it takes
is ``guts and imagination''! [A memorable phrase from
N.C. Chaitin's 1962 film The Small Hours.]